In the rapidly advancing world of data-driven decision-making, algorithms hold tremendous promise for organizations, but their effectiveness depends on how they are applied. A recent study, “Decision Authority and the Returns to Algorithms,” by Hyunjin Kim, professor at INSEAD Business School, Edward L. Glaeser, professor at Harvard University, Andrew Hillis, head of data science and underwriting at Parafin, Scott Duke Kominers, professor at Harvard Business School and co-principal investigator at D^3’s Crypto, Fintech, & Web3 Lab, and Michael Luca professor at Johns Hopkins, explores this dynamic, revealing critical insights around the interaction between algorithms and human decision-making authority, specifically within an Inspectional Services Department. This article summarizes the study’s key findings, helping business leaders understand how to maximize algorithmic value while respecting human expertise.
Key Insight: The Power of Simple Data
“The greatest gains come from integrating data into the decision process in the form of simple heuristics.” [1]
Hyunjin Kim and her research team found that even basic data integration could dramatically improve decision-making outcomes. Their study demonstrated that simple algorithms based on historical data significantly outperformed human decision-making in predicting restaurant health code violations. While many organizations focus on sophisticated models, this research suggests that complex algorithms are not always necessary.
Key Insight: Misalignment of Organizational Goals and Algorithmic Insights
“We also found that most departments placed high value on allocating full decision authority to inspectors. All departments that we interviewed gave inspectors ultimate discretion in prioritizing inspections, with 69% of departments rating inspector discretion as being very important.” [2]
One of the critical factors influencing the value organizations derive from algorithms is how decision authority is structured. Kim and her team highlight that decision-makers often retain significant discretion over whether to follow algorithmic recommendations, and this discretion frequently reduces the potential gains from predictive analytics. Although the study shows that algorithms consistently outperformed human decision-makers, inspectors frequently overrode these recommendations, diminishing the efficiency and effectiveness of the inspections.
Indeed, the study found that inspectors often disregarded algorithmic suggestions in favor of organizational objectives, such as minimizing travel time or focusing on overdue inspections. This approach rarely improved outcomes and highlights the disconnect between managerial and organizational priorities and algorithmic capabilities. The lesson here is that organizations need to align their strategic goals with algorithm outputs to fully benefit from predictive insights.
Key Insight: The Stakes of Getting It Right
“Our findings suggest that managing decision authority is an important consideration when seeking to use algorithms as decision aids. As firms increasingly make investments in data and AI, estimated at over $40 billion USD in 2020 and projected to double in the next few years, our findings offer relevant practical implications.” [3]
The research highlights that while firms are investing heavily in AI and algorithms, the real value comes from managing how decision-makers use these tools. Algorithms can enhance predictive accuracy, but without guidelines, human discretion can diminish these gains. To maximize the value of data-driven tools, managers should focus on creating clear guidelines for when and how decision-makers rely on algorithmic insights, balancing human judgment with the benefits of predictive analytics to ensure discretion is only applied when it genuinely enhances decision quality. Managers should also keep in mind potential limiting factors to decision makers’ implementation of algorithmic insights, such as private information unknown to the algorithm and potential individual aversions to the use of new technology.
Why This Matters
For business professionals, the insights from this study are highly relevant for improving decision-making processes. By understanding the balance between human judgment and algorithmic recommendations, organizations can unlock greater value from their data investments. Decision-makers must align their goals with the capabilities of their algorithms and structure decision authority in a way that encourages the use of data-driven insights. This approach ensures that both technology and human expertise are used to their full potential, driving better business outcomes.
References
[1] Hyunjin Kim, Edward L. Glaeser, Andrew Hillis, Scott Duke Kominers, and Michael Luca, “Decision Authority and the Returns to Algorithms,” Strategic Management Journal (January 23, 2024): 619-648, 621.
[2] Kim, et al., “Decision Authority and the Returns to Algorithms,” 643.
[3] Kim, et al., “Decision Authority and the Returns to Algorithms,” 622.
Meet the Authors
Hyunjin Kim is a professor in the Strategy area at INSEAD Business School. She researches how firms can manage data and algorithms to improve their strategic decision-making, and how these technologies change how firms compete and build competitive advantage. She earned her bachelor’s and doctoral degrees from Harvard University, and an M.Sc from the University of Oxford and the London School of Economics.
Edward Glaeser is the Fred and Eleanor Glimp Professor of Economics at Harvard University. He also leads the Urban Economics Working Group at the National Bureau of Economics Research, co-leads the Cities Programme of the International Growth Centre, and co-edits the Journal of Urban Economics. He received his A.B. from Princeton University in 1988 and his Ph.D. in Economics from the University of Chicago in 1992.
Andrew Hillis is the head of data science and underwriting at Parafin, which provides end to end infrastructure for platforms and allows them to offer business financing products to their small business service providers. He received his PhD in Economics from Harvard University.
Scott Duke Kominers is a Professor of Business Administration in the Entrepreneurial Management Unit, Co-Principal Investigator of D^3’s Crypto, Fintech and Web3 Lab, a Faculty Affiliate of the Harvard Department of Economics and of the Harvard Center of Mathematical Sciences and Applications. His first book is The Everything Token: How NFTs and Web3 Will Transform the Way We Buy, Sell, and Create.
Michael Luca is a professor and the director of the Technology and Society Initiative at the Johns Hopkins University, Carey Business School, and a faculty research fellow at the NBER. His research, teaching, and advisory work focuses on the design of online platforms, and on the ways in which data can inform managerial and policy decisions.